Instructions to use imabedalghafer/maqsm_model_pii_removal_gemma34b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use imabedalghafer/maqsm_model_pii_removal_gemma34b with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="imabedalghafer/maqsm_model_pii_removal_gemma34b")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("imabedalghafer/maqsm_model_pii_removal_gemma34b")
model = AutoModelForMultimodalLM.from_pretrained("imabedalghafer/maqsm_model_pii_removal_gemma34b", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use imabedalghafer/maqsm_model_pii_removal_gemma34b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "imabedalghafer/maqsm_model_pii_removal_gemma34b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "imabedalghafer/maqsm_model_pii_removal_gemma34b",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/imabedalghafer/maqsm_model_pii_removal_gemma34b

SGLang

How to use imabedalghafer/maqsm_model_pii_removal_gemma34b with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "imabedalghafer/maqsm_model_pii_removal_gemma34b" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "imabedalghafer/maqsm_model_pii_removal_gemma34b",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "imabedalghafer/maqsm_model_pii_removal_gemma34b" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "imabedalghafer/maqsm_model_pii_removal_gemma34b",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use imabedalghafer/maqsm_model_pii_removal_gemma34b with Docker Model Runner:
```
docker model run hf.co/imabedalghafer/maqsm_model_pii_removal_gemma34b
```

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

MAQSM PII Detection Model

Model Description

This model is a fine-tuned version of gemma3 specifically designed for Personally Identifiable Information (PII) detection and masking in Arabic and English text. The model identifies and replaces PII with [MASKED] tokens while preserving the original text structure and meaning.

Model Details

Base Model: google/gemma-3
Model Type: Causal Language Model
Fine-tuning Method: LoRA (Low-Rank Adaptation)
Languages: Arabic, English (Mixed-language support)
Task: PII Detection and Masking
License: Same as base model

Training Details

Training Data

Custom dataset for PII detection and masking
Mixed Arabic-English text samples
Various PII types including names, emails, phone numbers, addresses, IP addresses, national IDs, and financial information

Supported PII Types

The model is trained to detect and mask the following types of PII:

Personal Names: First, middle, and family names
Contact Information: Phone numbers, email addresses
Addresses: Physical addresses and locations
Identification Numbers: National IDs, SSNs, account numbers
Financial Information: Bank account details, credit card numbers
Technical Identifiers: IP addresses, MAC addresses
Dates: Birth dates and other personal dates
Cryptocurrency: Wallet addresses and transaction IDs

Usage

Quick Start

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline

# Load model and tokenizer
model_name = "imabedalghafer/maqsm_model_pii_removal_gemma34b"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype=torch.auto,
    device_map="auto",
    trust_remote_code=True
)

# Create pipeline
pipe = pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
    device_map="auto"
)

# System message
system_msg = """
**Role Definition**
You are a specialized language model trained to identify and mask Personally Identifiable Information (PII) within textual data. 
"""

# Example usage
def mask_pii(text):
    prompt = [
        {"role": "system", "content": system_msg},
        {"role": "user", "content": f"{text}"}
    ]
    
    generation = pipe(prompt, do_sample=True, temperature=0.7, top_p=0.8, top_k=20)
    output = generation[0]['generated_text'][-1]['content'].strip()
    
    
    return output

# Test the model
test_text = "اسمي أحمد محمد ورقم هاتفي 123-456-7890"
masked_text = mask_pii(test_text)
print(f"Original: {test_text}")
print(f"Masked: {masked_text}")

Expected Output Format

Input:

"اسمي أحمد محمد ورقم هاتفي 123-456-7890"

Output:

"اسمي [MASKED] ورقم هاتفي [MASKED]"

Limitations

The model is primarily trained on Arabic and English text
Performance may vary with domain-specific PII types not covered in training
May occasionally over-mask or under-mask certain edge cases
Requires careful prompt engineering for optimal results

Ethical Considerations

This model is designed to protect individual privacy by identifying and masking PII. However:

Users should verify masked output for completeness
The model should not be the sole method for PII protection in critical applications
Consider human review for sensitive documents
Be aware of potential false positives/negatives

Citation

If you use this model in your research, please cite:

@misc{maqsm_pii_model_2025,
  title={MAQSM PII Detection Model: Fine-tuned Gemma-3-4B for Arabic-English PII Masking},
  author={TheConsultants Team},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/imabedalghafer/maqsm_model_pii_removal_gemma34b}
}

License

This model inherits the license from the base model (google/gemma-3-4b-it). Please refer to the base model's license for terms and conditions.

Contact

For questions, issues, or collaboration opportunities, please reach out through the Hugging Face model repository or create an issue in the discussions section.

Disclaimer: This model is provided as-is for research and development purposes. Users are responsible for ensuring compliance with applicable privacy laws and regulations when using this model in production environments.

Downloads last month: -

Safetensors

Model size

5B params

Tensor type

BF16